AI Endoscopic Data
Research type
Research Study
Full title
Video Data Collection to train AI and develop AI enhanced Videolaryngoscopy Software
IRAS ID
343669
Contact name
Rajinder Chaggar
Contact email
Sponsor organisation
London North West University Healthcare NHS Trust
Duration of Study in the UK
1 years, 0 months, 1 days
Research summary
Background
A patient may require a breathing tube (tracheal tube) to be placed into their windpipe (trachea). This procedure is known as tracheal intubation and allows a breathing machine (ventilator) to move an oxygen-air mixture in and out of the patient’s lungs. Instruments called videolaryngoscopes are used to aid insertion of the tracheal tube through the patients mouth. Videolaryngoscopes come in different shapes and transmit a video image of the tracheal opening to a screen.
No equipment exists that can assist in identifying anatomy during intubation, manipulating the videolaryngoscope or delivery of the tracheal tube. The consequences of not being able to intubate or deliver oxygen could be catastrophic leading to death if no alternative strategy is used. The development of software/artificial intelligence assistance that support intubation using a videolaryngoscope would increase safety and reduce potential harm to the patient.
We have developed an AI assistance device that can support intubation using a Macintosh-shaped videolaryngoscope (this is a particular defined curve for a videolaryngoscope) in a manikin and patients. This was produced by collecting video data of Macintosh-shaped videolaryngoscope intubations and training the AI software to assist in intubation. Our software has been tested in and shown to be superior when compared to standard care (submitted for publication in a peer-reviewed journal, the British Journal of Anaesthesia). The patient trial is currently being ongoing at Inselspital in Bern, Switzerland using a CE certified, clinical version of larynGuide.Aim
We now wish to develop a similar AI assistance software that works with other types of videolaryngoscope including hyperangulated (very curved), straight, and varying sizes of Macintosh-shaped videolaryngoscopes. This would allow a complete ‘assistance’ package around the use of videolaryngoscopes regardless of which shaped device was used. There is NO difference to the usual care of patients. In order to develop the AI assistance we need to collect video data of the intubation using a videolaryngoscope so we can train the AI software. This video data can be recorded through the videolaryngoscope device itself onto a memory card and is anonymous. In addition to the video data we would need patient demographics including age, sex, ethnic background, height, weight and a comment about intubation using the video classification of intubation score. This information is necessary for undertaking the procedure and documentation, as standard care, and does not require additional tasks. We hope this will enable all patient groups to be able to benefit from the AI assistance device.REC name
Yorkshire & The Humber - Leeds West Research Ethics Committee
REC reference
25/YH/0079
Date of REC Opinion
7 Apr 2025
REC opinion
Favourable Opinion